Load-dependent extracellular matrix organization in ...kheradvar.eng.uci.edu/files/Alavi_et_al_AJP.pdfCALL FOR PAPERS Cardiovascular Responses to Environmental Stress Load-dependent

CALL FOR PAPERS Cardiovascular Responses to Environmental Stress

Load-dependent extracellular matrix organization in atrioventricular heartvalves: differences and similarities

S. Hamed Alavi,1,2 Aditi Sinha,1,2 Earl Steward,3 Jeffrey C. Milliken,3 and Arash Kheradvar1,2

1The Edwards Lifesciences Center for Advanced Cardiovascular Technology, University of California, Irvine, Irvine,California; 2Department Biomedical Engineering, University of California, Irvine, Irvine, California; and 3Division ofCardiothoracic Surgery, University of California, Irvine, Irvine, California

Submitted 9 March 2015; accepted in final form 17 May 2015

Alavi SH, Sinha A, Steward E, Milliken JC, Kheradvar A.Load-dependent extracellular matrix organization in atrioventricularheart valves: differences and similarities. Am J Physiol Heart CircPhysiol 309: H276–H284, 2015. First published May 22, 2015;doi:10.1152/ajpheart.00164.2015.—The extracellular matrix of theatrioventricular (AV) valves’ leaflets has a key role in the ability of thesevalves to properly remodel in response to constantly varying physiolog-ical loads. While the loading on mitral and tricuspid valves is significantlydifferent, no information is available on how collagen fibers change theirorientation in response to these loads. This study delineates the effect ofphysiological loading on AV valves’ leaflets microstructures using Sec-ond Harmonic Generation (SHG) microscopy. Fresh natural porcinetricuspid and mitral valves’ leaflets (n � 12/valve type) were cut andprepared for the experiments. Histology and immunohistochemistry wereperformed to compare the microstructural differences between the valves.The specimens were imaged live during the relaxed, loading, and un-loading phases using SHG microscopy. The images were analyzed withFourier decomposition to mathematically seek changes in collagen fiberorientation. Despite the similarities in both AV valves as seen in thehistology and immunohistochemistry data, the microstructural arrange-ment, especially the collagen fiber distribution and orientation in thestress-free condition, were found to be different. Uniaxial loading wasdependent on the arrangement of the fibers in their relaxed mode, whichled the fibers to reorient in-line with the load throughout the depth of themitral leaflet but only to reorient in-line with the load in deeper layers ofthe tricuspid leaflet. Biaxial loading arranged the fibers in between thetwo principal axes of the stresses independently from their relaxed states.Unlike previous findings, this study concludes that the AV valves’three-dimensional extracellular fiber arrangement is significantly differ-ent in their stress-free and uniaxially loaded states; however, fiberrearrangement in response to the biaxial loading remains similar.

collagen fibers; mechanics; uniaxial load; biaxial load

NEW & NOTEWORTHY

This study delineates the effect of physiological loading on atrio-ventricular heart valves’ leaflets microstructures using SecondHarmonic Generation microscopy. The presented data help un-derstanding the biomechanical responses of the AV valves to theload, and facilitate the development of more accurate and sophis-ticated constitutive models for native valve leaflets.

THE MITRAL AND TRICUSPID HEART valves, referred to as atrio-ventricular (AV) valves, control the blood flow between theatria and ventricles of the heart. Regurgitation is the mostcommon type of AV valve disease with a 1.7% prevalence ofmitral regurgitation (MR), which is the highest among all typesof valvular disease in the United States (18, 34). In its primaryform, MR usually results from the valves leaflets’ pathologicalweakening and gross changes in their extracellular matrix(ECM) that are referred to as myxomatous degeneration (1, 17,22). Alternatively, functional MR leads to stiffer leaflets andabnormal ECM remodeling (23).

Contrary to mitral valve regurgitation, functional tricuspidregurgitation (TR) is the most common form of TR, and isoften associated with left-sided valve disease mainly due toMR (10, 25, 42). This shows that regardless of the valve type,the ECM characteristics of AV valves’ leaflets have a key rolein the pathophysiology of these valves. The composition,arrangement, and structural organization of AV valves’ ECMare influenced by the mechanics of the left and right heartduring development and maturation as well as by load depen-dence tissue remodeling (24, 38).

Mitral and tricuspid valves are anatomical equivalents ofeach other, and both possess annulus, leaflets, chordae tendi-nae, and papillary muscles. From an anatomical perspective,the tricuspid valve has a larger orifice with three thinner andmore translucent leaflets, whereas the mitral valve is bileafletand its anterior and posterior leaflets are relatively thicker.Both valves’ leaflets are composed of a smooth endocardiallayer consisting of endothelial cells and a fibrous skeleton withspongiosa and fibrosa layers that are mostly composed ofcollagen fibers (21, 38, 41). Mitral and tricuspid valves areshown to have their fibers arranged in radial and circumferen-tial directions (11, 29, 38).

The leaflets’ microstructure determines its nonlinear stress-strain (20, 35, 43) and anisotropic characteristics (9, 36, 43)that influence leaflet functionality. Collagen fibers, in particu-lar, play a crucial role in providing structural integrity towithstand the cyclic loads during systole and diastole byconstant stretching and relaxation. It is well-known that themechanical response to cyclic loads is determined by the abilityof collagen fibers to orient in the direction of the principal stressesapplied to the leaflet (15, 16, 40). This necessitates accuratecharacterization of the valvular ECM and its remodeling po-tential under loading conditions. While loading conditions onmitral and tricuspid valves are significantly different due to thepressure difference in the left and right ventricles, no informa-

Address for reprint requests and other correspondence: A. Kheradvar, TheEdwards Lifesciences Center for Advanced Cardiovascular Technology, Univ.of California, Irvine, 2410 Engineering Hall, Irvine, CA 92697 (e-mail:[email protected]).

Am J Physiol Heart Circ Physiol 309: H276–H284, 2015.First published May 22, 2015; doi:10.1152/ajpheart.00164.2015.

0363-6135/15 Copyright © 2015 the American Physiological Society http://www.ajpheart.orgH276

mailto:[email protected]

tion exists on how collagen fibers change orientation in re-sponse to these stresses. This study intends to delineate the AVvalves’ differences and similarities in collagen fiber orientationin response to mechanical loading conditions. The resultsshould help better understanding the AV valves’ pathophysi-ology.

MATERIALS AND METHODS

Sample Preparation

Fresh porcine mitral and tricuspid heart valves (n � 12/type ofvalve) were obtained at the University of California Irvine (UCI)Department of Surgery (The UCI Institutional Animal Care and UseCommittee no. 1999-1712) and stored at 4°C in phosphate-bufferedsaline with 2% mixture of antibiotics including penicillin, streptomy-cin, and amphotericin B (GIBCO, Carlsbad, CA). The valves’ leafletswere excised and prepared for Second Harmonic Generation (SHG)microscopy. The tissue segments remained hydrated before and dur-ing the microscopy using the above-mentioned solution.

Biaxial Mechanical Loading Setup

To provide real-time monitoring of biaxial loads applied over a thintissue segment mounted on the stage of a high-resolution nonlinearmicroscope, a biaxial mechanical loading device (Fig. 1A) was devel-oped as described in detail in our previous publication (5). In a fewwords, this lightweight device consists of four loading grips connectedto adapters for clamping the tissue, one stage insert, one platform, fourpulleys, four tension screws, and four force gauges. The entire devicewas placed on the motorized X–Y stage on a Zeiss LSM 510 MetaMultiphoton microscope (Carl Zeiss Microscopy). The force ex-tended on each grip was monitored in real time using four digitalforce gauges with 50 mN resolution (Mecmesin, West Sussex,UK). In a near-frictionless situation, small pulleys guide the thincables from grips to the corresponding force gauges. To achievetension control, small grips held the tissue specimen at all foursides of the tissue segments. Tension control was achieved byplacing a 10 –32 screw between each grip and its correspondingforce gauge in a way that screw rotation increased tension in thecable. The force reading was calibrated by applying an initialpretension with a typical value of 0.15 N.

Loading/Imaging Experiment

SHG images were acquired by a Zeiss LSM 510 Meta Multiphotonmicroscope. The microscope is equipped with a Ti:Sapphire, Chame-leon-Ultra (Coherent, Santa Clara, CA) femtosecond laser sourcetunable from 690 to 1,040 nm. SHG excitation and emission filtrationwas set to 900 and 450–465 nm, respectively. Experiments wereperformed on three different regions on the belly part (Fig. 1, B andC) of the mitral and tricuspid valve leaflets (n � 12/type of valve).

The radial direction of the leaflet was identified as the direction ofthe conventional x-axis in the x–y image plane (Fig. 1C). SHG imagestacks with the field of view of 225 � 225 �m were acquired for eachsegment. The imaging was performed from the surface of the leafletsup to 60 �m deep inside the tissue under four different loadingconditions. At the beginning of the experiment, the leaflets wereimaged under low initial tension to determine the collagen fiberorientation in the relaxed state, which was then used to normalize thechanges of the collagen fiber orientation under uniaxial and biaxialloading conditions. Following the baseline imaging, the leaflets weresubjected to uniaxial radial loading and biaxial loading. Any displace-ment of the leaflet tissue was avoided while applying the loadingconditions during imaging. The tensile forces of 1.4 and 0.5 N wereapplied to mitral and tricuspid valve leaflets, respectively. Thesevalues are equivalent to the tensile forces experienced by a native

leaflet in the porcine heart (19). Finally, the leaflets were unloaded andimaged again.

Image Analysis

SHG image stacks were processed to extract mean collagen bundledirection from each image. The algorithms were all previously devel-oped and validated in-house using a MATLAB code (Mathworks)implementing two-dimensional (2D) Fourier Transform analysis (5,6). The frequencies of light-intensity oscillation for pixels werecalculated and rearranged to bring the zero frequency to the center ofthe image. An intensity threshold was defined as the mean value plusthree SDs to suppress the background of Fourier images by thresh-olding (7). The frequency indexes of the filtered images were ex-tracted and plotted. To calculate the mean fiber direction, a regressionline was defined through the distribution of frequencies, and then fiberdirection was extracted orthogonal to this trend. The Fourier Trans-form allows characterization of the frequency of light-intensity fluc-tuations for each pixel. Thus, it was possible to obtain the distributionof frequencies related to the spacing of the collagen fibers.

Fig. 1. The custom-made biaxial testing device mounted on the stage of astandard multiphoton microscope. A: the biaxial system was designed to applycontrolled uni- and biaxial tensile stresses to the tissue segments simultaneouswith the imaging. The lightweight system fits into the X–Y scanning stage ofthe microscope, and through using a pulley system the exerted force on eachside of the specimen is transferred to the force gauges. The magnitude of theforces is controlled using four screws placed between the grips connected tothe specimen and the force gauges. B: the sample tissue placed on the stage ofthe microscope-compatible biaxial tester. The tissue was placed in a specificorientation so that the uniaxial radial and biaxial tensile forces can be appliedin accordance of the radial and circumferential directions of the leaflet. C:schematic representation that demonstrates the radial and circumferentialdirections and the calculated fiber orientation angle on a sample leafletdrawing. The imaging experiment was performed at three different regions inthe belly of the leaflet that are shown with red triangles. The tensile forces of1.4 and 0.5 N were applied to multiple samples of mitral and tricuspid valveleaflets, respectively.

H277ECM ORGANIZATION IN ATRIOVENTRICULAR VALVES

AJP-Heart Circ Physiol • doi:10.1152/ajpheart.00164.2015 • www.ajpheart.org

Histology and Immunohistochemistry

Histology was performed by placing the leaflets in formalin.Paraffinized sections (5 �m) were stained by hematoxylin and eosin(H&E) as well as Masson’s trichrome for general morphology andECM collagen components, respectively. To assess the cellular phe-notypes, immunohistochemistry was performed by incubation withmonoclonal mouse antibodies for �-smooth muscle actin (�-SMA),vimentin, and CD31. DAB Chromogen substrate was used withMayer’s hematoxylin counterstain. A secondary biotin-labeled goatanti-mouse IgG antibody was used for incubation before the signaldeveloped. Normal light microscopy was used to analyze all thestaining.

Statistical Analysis

The data were derived from the SHG images in triplets and at threedifferent regions of the belly of the leaflet tissues (Fig. 1C); they arereported as means � SD. A total of 24 valves were used (n � 12/valvetype); therefore, 36 series of images and 108 series of data for mitralvalves and the same number for tricuspid valves were obtained. Anunpaired Student’s t-test was performed for statistical analysisusing the R software package for Windows (Lucent Technologies,Costa Mesa, CA). A P value of �0.05 was considered statisticallysignificant.

RESULTS

Histological Studies

The results for histology and immunohistochemistry on bothmitral and tricuspid valves are presented in Fig. 2. H&Estaining shows an organized structure for both valves withuniform cell densities (Fig. 2, A and B), whereas trichromestaining (Fig. 2, C and D) demonstrates a distinct collagenousfibrosa layer in tricuspid compared with mitral. In mitralleaflet, the collagen fibers were found abundantly present atalmost all layers; however, in tricuspid leaflets the collagenconcentration was found much higher at the fibrosa comparedwith the spongiosa layer. Immunohistochemistry data werefound quite similar in both valves with low levels of �-SMA-positive cells (Fig. 2, E and F). These cells were not distributedevenly in both valves. However, they tend to be more distrib-uted in the spongiosa layer rather than fibrosa. Endotheliallining was observed by CD31 staining (Fig. 2, G and H), andthe level of vimentin-positive cells was reported high in bothvalves (Fig. 2, I and J).

Loading-Imaging Studies

Mitral valve. Figure 3 shows the matrix map of a mitralanterior leaflet at the 10-, 40-, and 60-�m depths throughoutthe tissue before loading and under uniaxial radial and biaxialloading regimes. The ECM map of the unloading states ap-peared identical to the relaxed states (SHG data not shown). Avarying configuration of the fibers for the relaxed state throughthe depth in the leaflet’s belly can be observed. The fibers tendto arrange almost radially at the superficial layers (i.e., �30�m depth) and circumferentially at the deeper layers (i.e.,deeper than 30 �m). Once loaded radially, the fibers reorientdensely along with the load at all depths. The collagen bundlesalso become thicker compared with the relaxed state but notfully stretched. This is somehow different when the leafletswere subjected to biaxial loading, where the fibers changed toa stretched and completely straight configuration, as can be

seen in Fig. 3; the fibers reoriented in between the twoprincipal axes of the loads.

The comparisons of angle of orientation vs. depth at whichthe images were taken are shown in Figs. 4 and 5 for uniaxialradial and biaxial loadings, respectively. Each figure reportsthe relaxed, loading, and unloading cases. The mean bundledirection in relaxed states for the superficial layers is 3.7 �

Fig. 2. Histology and immunohistochemistry of the native mitral and tricuspidvalve leaflets. The first column displays the results for the mitral valve and thesecond column for the tricuspid valve. Hematoxylin and eosin (H&E) (A andB) revealed similar structure with uniform cell density (scale bars: 200 �m),but Masson’s trichrome (C and D) showed different distribution for collagencomponents (scale bars: 400 �m). Immunohistochemistry for �-smooth mus-cle actin (�-SMA, E and F) with scale bars � 200 �m, CD31 (G and H) withscale bars � 100 �m, and vimentin (I and J) with scale bars � 200 �mdemonstrates quite similar data with low amounts of �-SMA (brown colorshown by arrows)- and high amounts of vimentin-positive cells. CD31-positive(endothelial) cells are also identified by arrows as the lining layer of theleaflets.

H278 ECM ORGANIZATION IN ATRIOVENTRICULAR VALVES


1.1°; however, there is a sudden change in the fiber orientationfor deeper layers where the fiber mean orientation is 65.7 �2.8°. Fourier analysis indicates that uniaxial radial loadingreorients all fibers between 4° and 10° with an average of 6.5 � 0.9°,whereas biaxial loading reorients them between 45° and 51°with an average of 47.7 � 1.6°, almost in between the twoprincipal axes of the loads as observed in the SHG images.Unloading rapidly redirected the fibers to a position similar totheir relaxed state for both uniaxial and biaxial cases with P �0.83 and 0.71, respectively.

Tricuspid valve. The ECM configuration of a fresh porcinetricuspid leaflet at the depths of 10, 40, and 60 �m in relaxed

and under uniaxial radial and biaxial loadings are shown in Fig. 6.SHG images of unloading states are not shown, since they weresimilar to the relaxed states. An unorganized configuration forthe fibers can be observed in the relaxed states; the superficialfibers, in contrast to the ones observed in mitral leaflets, tend tobe more directed in between the radial and circumferentialdirections. They slightly turn toward the circumferential direc-tion at deeper layers, as also observed in the mitral leaflets.However, the uniaxial radial response was found to be differentcompared with mitral leaflet with collagen fibers aligning withthe radial load only at deeper layers. The superficial layersseem unaffected under the load by remaining at their relaxed

Fig. 3. Collagen fiber distributions of the mitral valveleaflet taken by loading-imaging technique described. Itdisplays the changes in the orientation of collagen fibersfor relaxed, uniaxial radial, and biaxial loadings at 10,40, and 60 �m deep inside the tissue. The unloadingimages have not been shown. As can be seen, collagenfiber orientation changes profoundly during the loadingcondition.

Fig. 4. Comparison of collagen fiber orientation in relaxed, uniaxial radialloading, and unloading states of the mitral valve leaflet. Uniaxial radial loadingshows that fibers are oriented in-line with the direction of the load.

Fig. 5. Comparison of collagen fiber orientation in relaxed, biaxial loading, andunloading states of the mitral valve leaflet. Biaxial loading shows that fibers areoriented in between the two principal axes of the loads.



original configuration, whereas the biaxial response was foundto be similar to the mitral valve with fibers aligning in betweenthe two principal axes of the stress. The types of fibers clearlylook different with a thicker and straighter shape in mitral vs.a wavier form with little curls in tricuspid (Figs. 3 and 6).

Figures 7 and 8 illustrate the average fiber orientation anglevs. the depth of imaging for uniaxial radial and biaxial load-ings, respectively. Each figure compares the data acquired forrelaxed, loading, and unloading conditions. The relaxed statesshow a variation of mean fiber direction from 23° to 81°throughout the tissue depth. This demonstrates that the fibers

are scattered in a smaller angle (�58°) once compared with themitral leaflet (�83°). Uniaxial radial loading slightly turns thesuperficial fibers �4° toward the circumferential direction(�5° for mitral leaflet) and reorient them at 29.8 � 2.0°(compared with 6.8 � 0.7° for mitral). Deeper inside the tissue,the fibers stand in between 6° and 1° with an average of 3.6 �0.4°(compared with 6.1 � 0.6° for mitral), almost aligned withthe radial load. This shows that the radial force acts similarlyin both AV valves and tends to make the deep fibers alignedwith the force while turning the superficial fibers slightlytoward the circumferential direction. Still, only in mitral leaf-lets and not tricuspids, the fibers are redirected all aligned with

Fig. 6. Collagen fiber distributions of the tricuspid valveleaflet taken by loading-imaging technique described. Itdisplays the changes in the orientation of collagen fibersfor relaxed, uniaxial radial, and biaxial loadings at 10,40, and 60 �m deep inside the tissue. The unloadingimages have not been shown. As can be seen, collagenfiber orientation changes profoundly during the loadingcondition.

Fig. 7. Comparison of collagen fiber orientation in relaxed, uniaxial radialloading, and unloading states of the tricuspid valve leaflet. Uniaxial radialloading shows that fibers are oriented in-line with the direction of the load onlyat deeper layers but not at the surface.

Fig. 8. Comparison of collagen fiber orientation in relaxed, biaxial loading, andunloading states of the tricuspid valve leaflet. Biaxial loading shows that fibersare oriented in between the two principal axes of the loads.



the load. The biaxial response is analogous (P � 0.08) to themitral leaflet with fibers reorienting in between 42° and 48°with an average of 45.3 � 1.9°. Unloading cases return thefibers to a position similar to their relaxed states for both uni-and biaxial conditions with P values of �0.79 and 1.05,respectively. This information has been summarized in Table 1to compare the microstructure of mitral and tricuspid valves atdifferent conditions.

DISCUSSION

Studying the heart valves’ ECM remodeling is crucial forunderstanding their physiological functions, their modes offailure, and for identifying congenital abnormalities duringtheir development and function. This knowledge would alsohelp in designing better prosthetic or tissue-engineered heartvalves. There are several previous studies describing the role ofcollagen fiber architecture in either a native or a bioprostheticheart valve (9, 12–14, 44). However, it is not known how thesefibers change their spatial orientation under physiological cy-clic loading. In other words, microstructural changes in colla-gen fibers during valve function in vivo are not well studied.These adjustments are extremely important in defining thebiomechanical response of the tissue to the varying degree ofstresses in a cardiac cycle. Any disruption of the extracellularorganization of the leaflets, especially collagen fiber orienta-tion, during development or later in adulthood due to a diseaseprocess may lead to structural and morphological abnormalitiessuch as thickening, stiffening, or valve leakage. In this study,for the first time, we investigated the temporal and spatialchanges in the pattern of the collagen fibers in native AVleaflets based on a novel simultaneous loading-imaging tech-nique.

The AV valves are similar in their gross anatomy andmicroscopic histology, yet they are subject to different me-chanical loading environments (38). During systole when theventricular pressure reaches up to or greater than 100 mmHg inthe LV, the mitral valve remains closed and therefore experi-ences significant hydrostatic pressure that can be fairly mod-eled with a biaxial loading system. This pressure change ismuch less dramatic in the right ventricle with a systolicpressure range of �15 to �30 mmHg. These differences in the

hemodynamics of the left and right ventricles would affect theleaflets’ microstructures during development and maturation,allowing them to adapt to their environment. This study delin-eates these difference by assessing the AV leaflets’ micro-scopic histology and their fiber arrangement in their relaxedmode (both cases are considered nearly stress-free conditions)compared with the loaded modes.

The Stress-Free Condition

Although the histology and immunohistochemistry datashow minimal differences between the two valves, the tricus-pid valves’ leaflets have considerably nonhomogenous colla-gen distributions with a distinct and more collagenous fibrosalayer than mitral leaflets, which can affect their mechanicalresponse to the load (Fig. 2). Grashow et al. showed that themechanical responses of mitral leaflets are strain-independentand have minimal hysteresis (20). This nonlinear stress-strainrelationship has been reported by May-Newman and Yin (36)and Liao et al. (33). This strain independency has been ratio-nalized based on collagen fiber recruitment theories (31, 32).Reduced collagen concentration in a diseased mitral valve wasobserved by Kunzelman et al. and led to increased tissuestiffness as well as reduced coaptation (30). However, there iscurrently no study showing the differences between collagenfiber orientation in mitral leaflets with a lower concentrationgradient and tricuspid leaflets with a higher gradient, in termsof their functions and their mechanical responses to the load.Based on the rule of mixtures, it is anticipated that such acollagen distribution in mitral leaflets can ensure the properclosure of the valves and strengthen their resistance to higherlevels of pressure during systole. Additionally, our data suggestthat the collagen fibers in mitral valves’ leaflets are generallythicker but straighter than the ones observed in tricuspid valves(Figs. 3 and 6), which should enhance their mechanical resil-ience in accordance with the previous statement.

The SHG data showed a significantly different (P � 0.004)configuration of the fibers in the superficial layers of therelaxed modes. The fibers in the deeper layers seem to orient ina similar fashion (P � 0.93). No significant differences wereobserved between fiber orientations of the valves from differ-ent animals with P � 0.23 and 0.51 for the mitral and tricuspid

Table 1. Comparison of mitral and tricuspid valves in stress-free and loaded conditions

Sress-Free Condition Loaded Condition

Histology Immunohistochemistry Relaxed Uniaxial Biaxial

H&E Trichrome �-SMA CD31 Vimentin Superficial Deep Superficial Deep Superficial Deep

Mitralvalve

Organizedstructurewith uniformcell density

Fibrosa: richin collagencontentSpongiosa:loweramount ofcollagen

Radial Inclined-towardcircumferentialdirection

Radial Radial Inclined, inthemiddleof thetwo axes

Inclined, inthemiddleof thetwo axes

Tricuspidvalve

Organizedstructurewith uniformcell density

Fibrosa: richin collagencontentSpongiosa:poor incollagencontent

Inclined,towardradialdirection

Circumferential Inclined,towardradialdirection

Radial Inclined, inthemiddleof thetwo axes

Inclined, inthemiddleof thetwo axes



valves, respectively. During the relaxed state, the fibers arescattered in a smaller angle in the tricuspid valve than in themitral valve. This might be due to the similar flow profiles ofboth valves, but to lower velocities in the tricuspid (almosttwo-thirds), due to its larger orifice, which lowers the shearstress on the surface during valve opening (2, 46). The radialfiber orientation on the surface and the circumferential fiberdirection in deeper layers would give the mitral valve leafletenough compliance to resist the surface shear stresses that aredamaging to superficial layers of the leaflet. Therefore, thispredefined configuration of the fibers created during valvedevelopment and maturation is key to describing valves’ me-chanical responses to different hemodynamic conditions in theleft and right heart. This would also define the heterogeneousand anisotropic properties of the valves seen in other studies(36, 43).

The Loaded Condition

According to the results from the uniaxial radial loadingexperiment, both valves’ responses were similar in nature otherthan the fact that the collagen fibers in the mitral valve align inthe direction of the load regardless of their position while thefibers in the tricuspid valve only align with the load in deeperlayers. By comparing the loaded condition with the relaxedmode, we infer that the uniaxial radial load cannot significantlychange the orientation of the fibers in superficial layers; how-ever, it dramatically shifts the deeper layers’ fiber orientationto a direction almost perpendicular to their original direction.We found that the uniaxial loading results for both valves weredependent on the fibers’ orientation in their relaxed mode. Thisarrangement would give mitral leaflets better extensibility thatallows them to accommodate greater tensile loads than tricus-pid leaflet. Lower tensile stress due to lower hydrostatic pres-sure in the right ventricle would cause the fibers in the tricuspidvalve to not align with the unidirectional load at all layers. Thismay also change the shape of the collagen fibers in thetricuspid leaflet to be thinner than and not as straight as thefibers in the mitral leaflet (Figs. 3 and 6). There is also apossibility that the different mechanical responses in mitral andtricuspid valves are due to the differences in their specificcollagen type contents. However, because the dominant and theload-bearing type of collagen in heart valves is type I collagen,we believe this factor would not significantly affect the ECMorientation of these tissues under the load.

We observed that the biaxial responses of both valves arequite similar and independent from their relaxed modes withfibers standing in between the two principal axes of the stressesin all layers. This ideal situation mimics the closure of thevalve where the AV valves experience tensile forces underhydrostatic pressure of the right or left ventricles during end-diastole. However, when the blood flows and produces shearstress on the surface of the leaflets, the situation will change. Inthis case, the optimal response is varied depending on the typeof the valve and the loading environment, as observed in ourresults. Therefore, unlike previous thoughts, the differentiatingfactor between mitral and tricuspid valves is their mechanicalresponse to the load once they are at their opening state and notwhen they are closed. The unloading response in all thesamples confirmed that there is no viscoplastic behavior orpermanent change in fiber orientation, which is in accordance

with the work done by other groups (20, 43). There arestructural constitutive models for planar collagenous tissuessuch as the one developed by Sacks (39) that considers theexperimentally derived fiber orientation. However, in all thesemodels the spatial distribution of fibers is an important missingfactor. Incorporating the findings of the present study into theseconstitutive models should better portray the mechanical char-acteristics of the valve.

It should be mentioned that the microstructure of the valveleaflets is not composed of only collagen but elastin andglycoaminoglycans components (Fig. 9); however, the me-chanical response of the leaflet to the load is mainly affected bythe collagen fiber distribution and orientation. This means thatthe collagen can be considered as the main determinant of thevalve’s microstructure. The collagen-rich layers of both valvesare in the fibrosa layer that has the highest level of collagenconcentration, which, according to our study, was found almostsimilar in both valves. Moreover, the fibrosa layer is the mainload-bearing component of the leaflet and defines its mechan-ical characteristics (28, 37, 45), which makes it more critical tostudy its load-dependent behavior. Nevertheless, we observedthat there were minimal differences in the fiber orientation indifferent regions of the leaflets at the same depth where thecollagen concentration was found different. This may implythat, on an inner layer parallel to the leaflet’s surface, thedifferences in collagen density would not significantly influ-ence a change in fiber orientation as can be seen in Fig. 9.Although we did not test it, we anticipate that this statementcan also be extended to the layers perpendicular to the leafletsurface since the fibers’ direction would not change the indi-vidual fibers’ characteristics and the biomechanical response ofthe tissue as a whole. Due to the structure of AV valves’collagen fibers, their changes in fiber direction can be more ofa compensatory response to the type of the applied load and notto the collagen concentration. Therefore, we think that thecollagen fibers in the spongiosa layer in lower concentration(a.k.a. the fiber orientation in deeper layers) would still showsimilar fiber orientation under the load. Accordingly, it isanticipated that the viscoelastic mechanical properties of theselayers in both valves would be completely different from thefibrosa layer.

Impact on Tissue-Engineered Heart Valves

The present data can be used as a model in defining themicrostructural response of the leaflet tissue to uniaxial andbiaxial loads. By implementing these models, an optimal wayto engineer a valve can be described considering the three-dimensional (3D) fiber orientations. Almost all of the compu-tational and constitutive models use a 2D scheme for definingfiber orientation, which may not be accurate considering thespatial sensitivity of the fibers to the load. Here we anticipatethat, by implementing the optimal fiber orientation data, onecan improve bioprosthetic and tissue-engineered valves’ dura-bility and functionality. Our results indicate that the uniaxialresponse of the native leaflets in deeper layers is quite similarto the uniaxial response of the bovine pericardial leaflets usedin bioprosthetic valves (5). However, the superficial layers’fiber orientation, which is a key determinant of the leaflet’sresponse to shear stress during valve opening, is somehowdifferent in bovine pericardial tissues from their orientation in



native valves. The biaxial response of the native leaflets is alsodifferent from the response of pericardial tissues with fibersarranged �45° in native valves and �60° in pericardial tissues(5). Because in the circumferential direction valves are lessextensible than they are in the radial direction (8), it wasexpected to see in response to the biaxial load the native valvefibers reorient closer to the axis where stiffness is higher. Thisobservation might be due to the differences in the anisotropicproperties of native and fixed pericardial leaflets (43), whichcan be used as a reference for designing more durable biopros-thetic heart valves. The observed differences between thesuperficial and deeper collagen fibers suggest that the decellu-larization process of the heart valves for tissue engineeringpurposes (26) should be performed carefully not to damage thenative fiber configuration of the most superficial layers. Agentle but longer decellularization process to reach the deeperlayers would be particularly beneficial for mitral valves sincethe alignment of its fibers will significantly influence its re-sponse to shear stress during valve opening.

In conclusion, we have characterized the ECM organizationof the AV valves in both stress-free and loaded conditions. Inthe stress-free state, we observed that not only the collagendistribution is significantly different in mitral and tricuspidvalves, but their fiber orientation is also different in 3D. In theuniaxial loading state the fibers in both valves rearrangeddifferently in a configuration dependent on their relaxed states.The biaxial response of the valves was quite similar and foundindependent of their relaxed configuration. The present datawould help in understanding the biomechanical response of the

AV valves to the load to develop more accurate constitutivemodels and to design naturally comparable tissue-engineeredand bioprosthetic heart valves. We believe this study is a firststep in analysis of the pathological situations such as in mitralprolapse, myxomatous disease, and other structural AV valvediseases. This may also help in optimizing material propertiesused in prosthetic or tissue-engineered heart valves (3, 4, 27).

Limitations

Due to the limited depth penetration of multiphoton tech-niques within turbid tissues, optical sectioning deeper than 60�m was not possible. Additionally, there might have been aslight dislocation between the relaxed, loading, and unloadingphases, which we tried to minimize.

ACKNOWLEDGMENTS

We thank the histology laboratory at the University of California Irvine(UCI) School of Medicine for assisting us in performing the histology andimmunohistochemistry of the samples. We also thank the Sue and Bill GrossStem Cell Research Center at UCI for providing facilities to perform themultiphoton microscopies and the related experiments.

GRANTS

This work is partially supported by a grant from Children’s Heart Founda-tion and one from Edwards Lifesciences Foundation.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

Fig. 9. Schematic representation of a cross section of a heart valve leaflet shows the layered fashion of the tissue with all the extracellular matrix components.Elastin (brown), collagen (green), and glycoaminoglycans (GAGs, orange) are distributed in the fibrosa and spongiosal layers. The fibrosa layer is rich in collagenfibers and more suitable for studying the load-dependent tissue remodeling. The Second Harmonic Generation images at the top were taken at the same depth(60 �m) from the surface of tissue inside the fibrosa. These images display different collagen density but almost similar fiber orientation on a plane parallel tothe surface of the leaflet.



AUTHOR CONTRIBUTIONS

Author contributions: S.H.A., A.S., E.S., and J.C.M. performed experi-ments; S.H.A. analyzed data; S.H.A., A.S., and A.K. interpreted results ofexperiments; S.H.A. prepared figures; S.H.A. and A.S. drafted manuscript;S.H.A. and A.K. edited and revised manuscript; A.K. and S.H.A. conceptionand design of research; A.K. approved final version of manuscript.

REFERENCES

1. Akhtar S, Meek KM, James V. Ultrastructure abnormalities in pro-teoglycans, collagen fibrils, and elastic fibers in normal and myxomatousmitral valve chordae tendineae. Cardiovasc Pathol 8: 191–201, 1999.

2. Alam M, Wardell J, Andersson E, Samad BA, Nordlander R. Char-acteristics of mitral and tricuspid annular velocities determined by pulsedwave Doppler tissue imaging in healthy subjects. J Am Soc Echocardiogr12: 618–628, 1999.

3. Alavi SH, Groves EM, Kheradvar A. The effects of transcatheter valvecrimping on pericardial leaflets. Ann Thoracic Surg 97: 1260–1266, 2014.

4. Alavi SH, Kheradvar A. Metal mesh scaffold for tissue engineering ofmembranes. Tissue Eng 18: 293–301, 2011.

5. Alavi SH, Ruiz V, Krasieva T, Botvinick EL, Kheradvar A. Charac-terizing the collagen fiber orientation in pericardial leaflets under mechan-ical loading conditions. Ann Biomed Eng 41: 547–561, 2013.

6. Alavi SH, Sinha A, Kheradvar A. Matrix remodeling in native atrioven-tricular valves’ leaflets in response to mechanical loading (Abstract).Circulation 130: A12281, 2014.

7. Ambekar. Ramachandra Rao R, Mehta MR, Leithem S, ToussaintJKC. Fourier transform-second-harmonic generation imaging of collagenfibers in biological tissues. In: Biomedical Optics and 3-D Imaging.Miami, FL: Optical Society of America, 2010, p. BSuD63.

8. Balguid A, Rubbens MP, Mol A, Bank RA, Bogers AJ, van Kats JP,de Mol BA, Baaijens FP, Bouten CV. The role of collagen cross-links inbiomechanical behavior of human aortic heart valve leaflets–relevance fortissue engineering. Tissue Eng 13: 1501–1511, 2007.

9. Billiar KL, Sacks MS. Biaxial mechanical properties of the natural andglutaraldehyde treated aortic valve cusp-part I: experimental results. JBiomechan Eng 122: 23–30, 2000.

10. Carpentier A, Deloche A, Hanania G, Forman J, Sellier P, Piwnica A,Dubost C, McGoon D. Surgical management of acquired tricuspid valvedisease (Abstract). J Thorac Cardiovasc Surg 67: 53, 1974.

11. Cochran R, Kunzelman K, Chuong C, Sacks M, Eberhart R. Nonde-structive analysis of mitral valve collagen fiber orientation (Abstract).ASAIO J 37: M447, 1991.

12. De Hart J, Peters G, Schreurs P, Baaijens F. Collagen fibers reducestresses and stabilize motion of aortic valve leaflets during systole. JBiomechan 37: 303–311, 2004.

13. Doehring T, Kahelin M, Vesely I. Mesostructures of the aortic valve. JHeart Valve Dis 14: 679–686, 2005.

14. Driessen NJ, Boerboom RA, Huyghe JM, Bouten CV, Baaijens FP.Computational analyses of mechanically induced collagen fiber remodel-ing in the aortic heart valve. J Biomechan Eng 125: 549–557, 2003.

15. Driessen NJ, Cox MA, Bouten CV, Baaijens FP. Remodelling of theangular collagen fiber distribution in cardiovascular tissues. BiomechanModel Mechanobiol 7: 93–103, 2008.

16. Driessen NJB, Peters GWM, Huyghe JM, Bouten CVC, Baaijens FPT.Remodelling of continuously distributed collagen fibres in soft connectivetissues. J Biomechan 36: 1151–1158, 2003.

17. Freed LA, Levy D, Levine RA, Larson MG, Evans JC, Fuller DL,Lehman B, Benjamin EJ. Prevalence and clinical outcome of mitral-valve prolapse. N Engl J Med 341: 1–7, 1999.

18. Go AS, Mozaffarian D, Roger VL, Benjamin EJ, Berry JD, Blaha MJ,Dai S, Ford ES, Fox CS, Franco S. Heart disease and stroke statistics–2014 update: a report from the American Heart Association (Abstract).Circulation 129: e28, 2014.

19. Grande KJ, Cochran R, Reinhall P, Kunzelman K. Stress variations inthe human aortic root and valve: the role of anatomic asymmetry. AnnBiomed Eng 26: 534–545, 1998.

20. Grashow J, Yoganathan A, Sacks M. Biaixal stress-stretch behavior ofthe mitral valve anterior leaflet at physiologic strain rates. Ann BiomedEng 34: 315–325, 2006.

21. Gross L, Kugel M. Topographic anatomy and histology of the valves inthe human heart (Abstract). Am J Pathol 7: 445, 1931.

22. Hayek E, Gring CN, Griffin BP. Mitral valve prolapse. Lancet 365:507–518, 2005.

23. He S, Fontaine AA, Schwammenthal E, Yoganathan AP, Levine RA.Integrated mechanism for functional mitral regurgitation leaflet restrictionversus coapting force: in vitro studies. Circulation 96: 1826–1834, 1997.

24. Hinton RB, Yutzey KE. Heart valve structure and function in develop-ment and disease. Annu Rev Physiol 73: 29–46, 2011.

25. Izumi C, Iga K, Konishi T. Progression of isolated tricuspid regurgitationlate after mitral valve surgery for rheumatic mitral valve disease. J HeartValve Dis 11: 353–356, 2002.

26. Kheradvar A, Groves E, Dasi L, Alavi SH, Tranquillo R, Grande-Allen KJ, Simmons C, Griffith B, Falahatpisheh A, Goergen C,Mofrad MK, Baaijens F, Little S, Canic S. Emerging trends in heartvalve engineering. Part I. Solutions for future. Ann Biomed Engineer 43:833–843, 2015.

27. Kheradvar A, Groves E, Goergen C, Alavi SH, Tranquillo R, Sim-mons C, Dasi L, Grande-Allen KJ, Mofrad MK, Falahatpisheh A,Griffith B, Baaijens F, Little S, Canic S. Emerging trends in heart valveengineering. Part II. Novel and standard technologies for aortic valvereplacement. Ann Biomed Engineer 43: 844–857, 2015.

28. Kunzelman K, Cochran R, Murphree S, Ring W, Verrier E, EberhartR. Differential collagen distribution in the mitral valve and its influence onbiomechanical behaviour. J Heart Valve Dis 2: 236–244, 1993.

29. Kunzelman KS, Cochran R. Stress/strain characteristics of porcinemitral valve tissue: parallel versus perpendicular collagen orientation. JCard Surg 7: 71–78, 1992.

30. Kunzelman KS, Quick DW, Cochran RP. Altered collagen concentra-tion in mitral valve leaflets: biochemical and finite element analysis. AnnThoracic Surg 66: S198–S205, 1998.

31. Lanir Y. Constitutive equations for fibrous connective tissues. J Bio-mechan 16: 1–12, 1983.

32. Lanir Y. A structural theory for the homogeneous biaxial stress-strainrelationships in flat collagenous tissues. J Biomechan 12: 423–436, 1979.

33. Liao J, Yang L, Grashow J, Sacks MS. The relation between collagenfibril kinematics and mechanical properties in the mitral valve anteriorleaflet. J Biomechan Eng 129: 78–87, 2007.

34. Maganti K, Rigolin VH, Sarano ME, Bonow RO. Valvular heartdisease: diagnosis and management. In: Mayo Clinic Proceedings. NewYork, NY: Elsevier, 2010, p. 483–500.

35. May-Newman K, Yin F. A constitutive law for mitral valve tissue. JBiomechan Eng 120: 38–47, 1998.

36. May-Newman K, Yin FC. Biaxial mechanical behavior of excisedporcine mitral valve leaflets. Am J Physiol Heart Circ Physiol 269:H1319–H1327, 1995.

37. McCarthy KP, Ring L, Rana BS. Anatomy of the mitral valve: under-standing the mitral valve complex in mitral regurgitation. Eur J Echocar-diogr 11: i3–i9, 2010.

38. Misfeld M, Sievers HH. Heart valve macro-and microstructure. PhilTrans Royal Soc B Biol Sci 362: 1421–1436, 2007.

39. Sacks MS. Incorporation of experimentally-derived fiber orientation intoa structural constitutive model for planar collagenous tissues. J BiomechanEng 125: 280–287, 2003.

40. Sacks MS, Smith DB, Hiester ED. The aortic valve microstructure:effects of transvalvular pressure. J Biomed Materials Res 41: 131–141,1998.

41. Schoen FJ. Evolving concepts of cardiac valve dynamics: the continuumof development, functional structure, pathobiology, and tissue engineer-ing. Circulation 118: 1864–1880, 2008.

42. Spinner EM, Shannon P, Buice D, Jimenez JH, Veledar E, Pedro J,Adams DH, Yoganathan AP. In vitro characterization of the mechanismsresponsible for functional tricuspid regurgitation. Circulation 124: 920–929, 2011.

43. Stella JA, Liao J, Sacks MS. Time-dependent biaxial mechanical behav-ior of the aortic heart valve leaflet. J Biomechan 40: 3169–3177, 2007.

44. Tower TT, Tranquillo RT. Alignment maps of tissues. II. Fast harmonicanalysis for imaging. Biophys J 81: 2964–2971, 2001.

45. Vesely I, Noseworthy R. Micromechanics of the fibrosa and the ventric-ularis in aortic valve leaflets. J Biomechan 25: 101–113, 1992.

46. Yoganathan A. Heart valve dynamics. Biomechanics: Principles andapplications.



Documents

Load-dependent extracellular matrix organization in ...kheradvar.eng.uci.edu/files/Alavi_et_al_AJP.pdfCALL FOR PAPERS Cardiovascular Responses to Environmental Stress Load-dependent